LOCAL SCALE MAPPING OF NET PRIMARY PRODUCTION IN TROPICAL RAIN
FOREST USING MODIS SATELLITE DATA
Abd Wahid Rasib, Ab. Latif Ibrahim, A.P. Cracknell and Mohd Azahari Faidi
Department of Remote Sensing University Teknologi Malaysia, 81310, Skudai, Johor, MALAYSIAwahid@fksg.utm.my
KEY WORDS: Local scale, Mapping, Net Primary Production, MODIS, Tropical Rain Forest
ABSTRACT:
Satellite data from Moderate Resolution Imaging Radiometer (MODIS) is currently being used to extract net primary production
(NPP) at global scale. Over the years, small scale on-ground flux tower measurements using the eddy covariance method have been
used to validate the global MODIS NPP at a number of test sites. Nevertheless, in anticipation of local scale NPP, MODIS satellite
data is inadequately attempted to map NPP at plot size particularly in tropical rain forest region. This is due to the low spatial
resolution (250m to 1000m) of MODIS satellite data and the variability of NPP with locations. Thus, in this study, MODIS satellite
data is used to map local scale NPP for Pasoh Forest Reserve in Malaysia. Micrometeorological approach based on Monteith’s
equation is used to map NPP from MODIS satellite data for three years (2004, 2005 and 2006). The result shows that the pattern of
NPP concentration is slightly decrease from 2004 to 2006. NPP estimated using MODIS satellite data for Pasoh Forest Reserve are in
the range from 412.29 to 710.36 gCm2y-1, 392.16 to 684.96 gCm2y-1 and 411.19 to 631.65 gCm2y-1 for year 2004, 2005 and 2006,
respectively. These indicate that MODIS satellite data is appropriate to map local scale NPP of tropical rain forest.
1. INTRODUCTION
Recently, validation of annual global MODIS (Moderate
Resolution Imaging Radiometer) NPP (net primary
production) at 1 km spatial resolution posed
a great challenged and this issue has been seriously discussed
by many researchers (Turner et al., 2006). One of the major
issues that have been highlighted is regarding the
appropriateness of the variety of product scale such as to match
the low spatial resolution of MODIS satellite data with plotscale flux tower measurements on the ground (Cohen et al.,
2003; Turner et al., 2005). In addition, limited number of flux
tower worldwide is also contributing to this predicament. At
present, there are about 450 sites flux tower distributed
worldwide. Among these are AmeriFlux in United States,
Fluxnet-Canada in Canada, CarboEurope in Europe, AsiaFlux
in Asia, KoFlux in Korea, OzFlux in Australia, ChinaFlux in
China, and CarboAfrica in Africa. These have been seen as
insufficient for the validation of the global MODIS NPP, thus
cannot produced an accurate measurement of global NPP
(Vargas et al., 2007; Falge et al., 2001). There are also various
MODIS NPP models such as initiated by MOD17 (Running et.
al., 1999), “continuous field model” (Rahman et al., 2004) and
recent “Carnegie-Ames-Stanfor Approach model” (CASA)
(Jinguo et al., 2006) that employ either micrometeorological
approaches or biometric approaches have been developed
intensively to provide a consistent and continuous global
estimate of primary production (Heinsh et al., 2006). The
development of these series of model revealed that the
validation of annual MODIS NPP is an essential step which
provides a means of evaluating spatial patterns in productivity
as well as interannual variation and long term trends in
biosphere behavior. The main objective of this study is to map
local scale NPP for an area of about 600 hectares in tropical rain
forest low resolution MODIS satellite data. This study has been
carried out using the pioneer micrometeorological approach
according to the Monteith’s equation as follows:
NPP=LUE x APAR
(1)
(Monteith, 1977)
where, APAR refers to absorbed photosynthetically active
radiation and LUE is the light use efficiency. Equation (1) has
been modified by Rahman et al. (2004) to suit with the MODIS
images by analyzing the ability of utilizing MODPRI (MODIS
Photochemical Reflectance Index) to replace LUE. This new
equation is given as follows:
NPP= MODPRI x APAR
(2)
This new equation was then been used to map the NPP using
MODIS data for 2004, 2005 and 2006. Each image has been
sub-setted into specific local scale forest in Peninsular Malaysia
at Pasoh Forest Reserve whereby to enhance the main focus of
this study. However, due to the limitation in ability to acquire
field measurements data, results from this study has been verify
using the existing estimation of NPP values from compilation of
previous studies such as Clark et al., (2001); and Abd Wahid
Rasib et al., (2007). Results from this study can be used as
supportive sources to validate global MODIS NPP product as
well as to develop new carbon model in future.
2. STUDY SITE AND METHODS
This study has been conducted in one of the oldest tropical rain
forest, area the Pasoh Forest Reserve, located in Negeri
Sembilan, Peninsular Malaysia (Figure 1). It is situated between
latitude 2° 58’N and longitude 102° 18’E and its topography
1441
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
lies between 75m to 150m above sea level (World Geodetic
System 1984). The core area of about 600 hectare of Pasoh
Forest Reserve is a primary lowland mixed dipterocarp forest
(tropical evergreen broad-leaved forest) with terrain type is
quoted as gentle hill slope. This forest reserve is surrounded by
various species of tree such as Dipterocarpaceae, Leguminosae,
and Burseraceae as well as Euphorbiaceae, and Annonaceae for
dominant understory species. The continuous canopy height is
approximately 35 m, although some emergent trees exceed 45 m.
Pasoh Forest Reserve generally receives relatively high average
annual rainfall of 1733mm (2003-2005). The highest rainfall
normally occurs during the months of March to May and
September to January. Meanwhile, the mean monthly
temperature is in the ranges of 24.5°C to 27°C (2002-2005).
This Forest Reserve contains three experimental plots that have
been used extensively for studying primary productivity of
tropical rainforest during International Biological Program (IBP)
since 1978 (Kira 1978). Currently there are still some on-going
joint research project between Forest Research Institute
Malaysia (FRIM) and foreign research center (i.e. National
Institute for Environmental Studies, Japan – NIES) which
emphasize to the NPP study from remote sensing (Ab. Latif
Ibrahim and Okuda, 2005). There are also some others related
research been conducted using tower flux measurements and
biomass measurements that provide useful information for this
study (Abd Rahim Nik et al., 2005 and Abd. Rahman Kassim et
al., 2005). Moreover, there are some sophisticated instruments
that are currently located on the top of flux tower (52 meter
height) which is located in about 6 hectares plot such as LI190
(LICOR, USA) for PAR and LUE observation, HMP45
(VAISALA, Finland) for air temperature recorded and RT-5
(Ikeda
keiki,
Japan)
for
precipitation
observation
(http://asiaflux.net). Furthermore, the biometric parameter such
as DBH is well documented by FRIM from previous researches
in these experimental plots.
stress-induced reduction in NPP of terrestrial vegetation. This
model known as simple “continuous field model” solely based
on spectral data whereby significantly utilizing only the visible
and near infrared bands (band 1: 620-670 nm and band 2: 841876 nm) as well as the “ocean” bands (band 11: 526-536 nm
and band 12: 546-556 nm) to explain the variability of fluxtower based daily NPP. On the other hand, the most noteworthy
is a spectral index called the “Photochemical Reflectance
Index” (PRI) was determined and
has been analyzed
appropriately using tower-based LUE values. This satellitebased PRI has been proven for the ability to track the changes in
landscape-level photosynthesis activity. The definition of the
PRI is given by;
PRI = (ρ
531
- ρ )/(ρ
ref
531
+ρ )
(3)
ref
where, ρ represents reflectance at the wavelength (nm)
expressed by the numeral subscripts, and ref represents a
reference wavelength, typically 550 or 570 nm. Thus, the
calculation of MODIS-derived PRI (or MODPRI) is given as;
MODPRI = (ρ
where ρ
b11
and ρ
b11
b12
-ρ
b12
)/(ρ
b11
+ρ )
12
(4 )
represent band 11 and band 12, of MODIS
data respectively. In addition, the calculation of the Normalized
Difference Vegetation Index (NDVI) using band 1 and band 2 is
given as;
NDVI = (ρ - ρ ) / (ρ + ρ )
b2
b1
b2
b1
(5)
Subsequently, the NDVI values are utilized to calculate the
fraction of PAR (fPAR) absorbed by vegetation using the
following equations;
fPAR = 1.24 x NDVI – 0.168
Figure 1: Location of the study area
For the purpose of this study, level 1B of Terra MODIS satellite
data for Peninsular Malaysia has been acquired from
http://ladsweb.nascom.nasa.gov. Three years of
MODIS
satellite data dated 2th October 2004, 10th October 2005 and 2th
October 2006, respectively with minimum cloud cover have
been selected using quick-look menu provided in this free
download website. Each of the data has been geocoded to the
Malaysian Rectified Skew Orthomorpic (MRSO) projection
coordinate system.
Model developed by Rahman et al. (2004) based on
micrometeorological approach was employed to determine the
local scale NPP from MODIS data in this study. Rahman et al.
(2004) have revealed that their new model is capable to track
the changing photosynthetic light use efficiency (LUE) and
(6)
Therefore, absorbed PAR (APAR) for each pixel can be
calculated from the relationship: of APAR = fPAR x PAR. The
PAR values is actually restricted to just a portion of
electromagnetic spectrum from 0.4 to 0.7 micrometers (μm)
which is comparable to the range of light the human eye can see.
Therefore, this value is assumed to be approximately 0.5 of the
incoming solar radiation (Prasad et al. 2002). Consequently, the
model has revealed a linear relationship between NPP and the
(MODPRI x APAR), which can explain 88% of the temporal
variability in NPP. Finally in this model NPP can be estimated
using the following regression equation;
NPP = 0.5139 (MODPRI x APAR) – 1.9818
(7)
This equation generates the value of NPP in units of g Cm2 y-1.
1442
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
3. RESULTS AND DISCUSSION
Three years of local scale NPP (2004, 2005 and 2006) for an
area of 600 hectares or approximately 190 pixels of MODIS
satellite data for Pasoh Forest Reserve have been produced in
this study using micrometeorological approach. NDVI and NPP
for the whole of Peninsular Malaysia have been extracted as
well whereby to be used as a reference map in order to verify
the final output of this study. However, only NPP map for 2004,
2005 and 2006, respectively (Figure 2 (a), (b) and (c)) will be
used to determine the range of NPP concentration in Peninsular
Malaysia.
(a)
(a)
(b)
(b)
(c)
(c)
Figure 3: MODIS NDVI map for Pasoh Forest Reserve. (a)
2004; (b) 2005; (c) 2006.
(a)
Figure 2: MODIS NPP for Peninsular Malaysia (a) 2004;
2005; (c) 2006
(b)
Two sets of NDVI and NPP maps for Pasoh Forest Reserve
have been succesfully accomplished using composite of visible
and near infrared MODIS satellite data. Figure 3 (a), 3(b) and3
(c) are the NDVI maps showing pattern of greeness level at
Pasoh Forest Reserve. While, Figure 4 (a), 4 (b) and 4 (c)
display the NPP maps and these are the final output of this
study. All output in Figure 2, 3, and 4, can be further explained
statistically as in Table 1.
1443
(b)
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
As an overall the values of NPP from three years MODIS
images in this study are comparable to the others previous
similar researches. Goetz et al., (2000) by using the series of
satellite observation have determined that the global annual
NPP is ranged from 0 to 1500 g m2 y-1 (7.5 t ha-1yr-1). Running
et al., (2004) showed the range between 100 to 1500 g C m2 y1
for estimation global annual NPP using MODIS images. Study
by Clark et al., (2001) have estimated that the NPP values in
tropical rain forest base on biometric approach is between 170
to 2170 gCm2y-1. Previous study conducted by Abd. Wahid
Rasib et al., (2007) estimated that the average value of NPP in
2004 for a smaller scale of Pasoh Forest Reserve was 633.85
gCm2 y-1. Thus, output from this study have showed very
relevence range in order to present the concentration of NPP to
the very dense Pasoh Forest Reserve with limited field data
acquistion.
(c)
Figure 4: MODIS NPP map for Pasoh Forest Reserve. (a) 2004;
(b)2005; (c) 2006.
Area
Year
Peninsular
Malaysia
2004
2005
2006
2004
2005
2006
Pasoh
Forest
Reserve
NDVI
Min.
Max.
-0.60
0.79
-0.31
0.75
-0.23
0.74
0.20
0.73
0.10
0.48
0.11
0.47
2 -1
NPP (gCm y )
Min.
Max
39.69
740.99
141.98
719.51
202.52
673.66
412.29
710.36
392.16
684.94
411.19
631.65
Table 1: The range of NDVI and NPP for Peninsular Malaysia
and Pasoh Forest Reserve.
Three years NDVI and NPP maps in 2004, 2005 and 2006 for
Pasoh Forest Reserve are within the range NDVI and NPP for
Peninsular Malaysia. While, minimum values for NDVI and
NPP in Peninsular Malaysia is slightly increased from 2004 to
2006. However, maximum values show decrement. This pattern
has indicated that the range for minimum and maximum values
in Peninsular Malaysia during these three years has reduced.
The minimum values for NDVI and NPP in Pasoh Forest
Reserve were fluctuated from 2004 to 2006 but the maximum
values were decreased. In addition, histogram in Figure 5
indicated that the NPP concentration was frequently distributed
in the range of 550 to 680 gCm2y-1 during these three years for
Pasoh Forest Reserve.
4. CONCLUSION
This study has shown the used of series low resolution MODIS
images for estimation of NPP using micrometeorological
approach at local scale about 600 hectares area. Three years of
MODIS images for the month of October 2004, 2005 and 2006,
have been used to determine the temporal distribution NPP in
Pasoh Forest Reserve. The pattern of NPP concentration in
Pasoh Forest Reserve seems to be decreased from 2004 to 2006.
The NPP for Peninsular Malaysia also shows similar pattern.
Model developed by Rahman et al. (2004) using MODIS ocean
bands in 80-years old broadleaf deciduous forest in Southern
Indiana (39_190N, 86_240W, 275 m above sea level) also
called as “continuous field model” have been successfully
applied to the very huge and old local scale tropical rain forest
such as Pasoh Forest Reserve. This model is solely based on
remotely sensed spectral data that could explain 88% of
variability in flux-tower based daily NPP. Results from this
study can be used to provide very useful basic information
related to the forest productivity mapping in tropical rain forest
as well as to improve estimating of the spatial and temporal
NPP using current global-scale MOD17.
REFERENCES
Ab. Latif Ibrahim and Okuda, T., 2005. Mapping net primary
productivity of Peninsular Malaysia forest using satellite data.
Annual report of the NIES/FRIM/UPM joint reserch project on
tropical ecology and biodiversity 2005. pp: 111-118
Abd Rahman Kassim, Azizi Ripin, Nur Supardi Md. Noor,
Niiyama, K., Kajimoto, T. Matsuura, Y., and Yamashita, T.,
2005. Allometric relationship between stem diameter, tree
height, leaf, stem and root biomass in Pasoh Forest Reserve.
Annual report of the NIES/FRIM/UPM joint reserch project on
tropical ecology and biodiversity 2005. pp: 27-36
Abd Rahim Nik, Siti Aishah Shamsuddin, Ahmad Abdul Salam,
Mohd Md Sahat, Takanashi, S., Tani, M., Kosugi, Y., Matsuo,
N., Ohkubo, S., Fukui, Y., Nakanishi, R., Noguchi, S., and
Okuda, T, 2005. Understanding the roles of tropical forest in
climate change through the energy/H2O/CO2 exchange
processes. Annual report of the NIES/FRIM/UPM joint reserch
project on tropical ecology and biodiversity 2005. pp: 1-10
Figure 5: MODIS NPP distribution histograms for Pasoh Forest
Reserve in 2004, 2005 and 2006, respectively.
Abd Wahid Rasib, , Ab. Latif Ibrahim, , Mohd Azahari Fandi,
and Wan Hazli Wan Kadir, 2007. Mapping Net Primary
Production in Tropical Rain Forest Using MODIS Satellite
1444
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
Data. In: The 28th Asian Conference on Remote Sensing 2007,
12-16 November 2007, PWTC, Kuala Lumpur, Malaysia.
Science.
Vol.
16,
No.4/November.
http://www.springerlink.com/content/t5h8607m5t575183/
Clark, A. D., Brown, S., Kicklighter, W. D., Chambers, Q. J.
Thomlinson, R. J., Jian Ni and Holland A. E. 2001. Net primary
production in tropical forests: An evaluation and synthesis of
existing field data. Ecological Applications, 11(2). pp: 371-384.
Kira T. 1978. Primary productivity of Pasoh forest. Malayan
Nature Journal 30: pp:291-297
Cohen, W. B., Maiersperger, T. K., Yang, Z., Gower, S. T.,
Turner, D. P., Ritts, W. D., Berterretche, M. and Running, S. W.,
2003. Comparisons ofland cover and LAI estimates derived
from ETM+ and MODIS for four sites in North America: A
quality assessment of 2000/2001 provisional MODIS products.
Remote Sensing. Environment, Vol. 88, pp. 233–255.
Monteith, J. L., 1977. Climate and the efficiency of crop
production in Britain. Philisophical Transactions of the Royal
Society of London, Series B, (281). pp: 277-294.
Rahman, A. F., Cordova, D. V., Gamon, A. J., Schmid, P. H.,
and Sims, A. D. 2004. Potential of MODIS ocean bands for
estimating CO2 flux from terrestrial vegetation: A novel
Approach. Geophysical Research Letters. (31) L10503.
Falge, E., D. Baldocchi, R. Olson, P. Anthoni, M. Aubinet, C.
Bernhofer, G. Burba, G. Ceulemans, R. Clement, H. Dolman, A.
Granier, P. Gross, T. Grunwald, D., Hollinger, N. O. Jensen, G.
Katul, P. Keronen, A. Kowalski, C. T. Lai, B. E. Law, T.
Meyers, J. Moncrieff, E. Moors, J. W. Munger, K. Pilegaard, U.
Rannik, C., Rebmann, A. Suyker, J. Tenhunen, K. Tu, S. Verma,
T. Vesala, K. Wilson, and S. Wofsy. 2001. Gap filling strategies
for long term energy flux data sets. Agricultural and Forest
Meteorology 107:71-77.
Running W. S., Nemani, R., Heinsch, A. F, Maosheng Zhao,
Reevee M, and hirofumi hashimoto 2004. A continuous
satellite-derived Measure of global terrestrial primary
production. BioScience. Vol. 54. (6) • Pp 547-560
Goetz, J. S., Prince, S. D., Small, J., Gleason, A. and Thawley
M. M. 2000. Interannual variability of global terrestrial primary
production: Results of a model driven with satellite
observations. Journal of Geophysical Research. (105). pp:
20,077-20,091
Turner, D. P., Ritts, W. D., Zhao, M.,. Kurc, S. A., Dunn, A. L.,
Wofsy, S. C., Small, E. E.,and Running, S. W., 2006.
Assessing Interannual Variation in MODIS-Based Estimates of
Gross Primary Production. IEEE Transactions on Geoscience
and Remote Sensing, Vol. 44, No. 7.
Heinsh, F.A., M. Zhao, Running S.W., Kimball J.S., Nemani
R.R., Davis K.J., Bolstad P.V, Cook B.D., Desai A.R., Ricciuto
D.M., Law B.E., Oechel W.C., Kwon H., Luo H., Wofsy S.C.,
Dunn A.L., Munger J.W., Baldocchi D.D., Xu L., Hollinger
D.Y., Richardson A.D., Stoy P.C., Siqueira M.B.S., Manson
R.K., Burns S.P., and Flanagan L.B.. 2006. Evaluation of
remote sensing based terrestrial productivity from MODIS
using regional tower Eddy flux network observations. IEEE
Transactions on Geoscience and Remote Sensing. 44: 19081925.
Turner, D. P., Ritts, W. D., Cohen, W. B., Maeirsperger, T.,
Gower, S. T., Kirschbaum, A., Running, S. W. , Zhao, M.,
Wofsy, S. C., Dunn, A., Law, B. E., Campbell, J. L., Oechel,
W., Kwon, H., Meyers, T. P., Small, E. E., Kurc, S. A., and
Gamon, J. A., 2005. Site-level evaluation of satellite-based
global terrestrial gross primary production and net primary
productionmonitoring. Global Change Biology., vol. 11, pp.
666–684
Jinguo, Y., , Niu, Z., and Wang, C., 2006. Vegetation NPP
distribution based on MODIS data and CASA model—A case
study of northern Hebei Province. Earth and Environmental
Running, W. S., Nemani, R., Glassy, M. J., and Thornton, E.
P., 1999. MODIS Daily Photosynthesis (psn) and Annual Net
Primary Production (npp) product (MOD17) Algorithm
Theoretical Basis Document. Version 3.0. 29 April
Vargas, R., Cook, R., Agarwal, D., Boden, T., Papale, D. Van
Ingen, C., Yang, B., Baldocchi, D., 2007. The Future of the
Worldwide Flux Tower Databases. FLUXNET Workshop. Oak
Ridge National Laboratory. Oak Ridge, TN, USA. October 2324.
1445
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1446